Project II: This project will determine, for the first time, whether there are differences in how school and community characteristics are related to drug use rates in rural ethnic minority communities. The first part of the project is a secondary data analysis, combining existing data from two sources, (1) a survey of a stratified random sample of 180 rural communities across the U.S. and (2) a survey of 30 rural African American and 30 rural Mexican American communities. The surveys in these communities provide data on drug use by 7th through 12th grade students, as well as data on social and psychological risk factors for drug use and cultural identification. Information on school and community characteristics will be available from the census (including the past census and a census that will be run during the course of this project), from the School Data Book and from telephone interviews with key informants in each community. Except for the new census, the data collection for phase I, the secondary data analysis, will be complete by the time this project begins. Community characteristics include rurality and is9lation, poverty, education level and dropout, crime and violence, community instability, community cohesiveness and attachment, industrial structure, family disorganization, family cohesiveness and extended family, religious heterogeneity, ethnic heterogeneity, age structure, and gender normals and roles. Hierarchical linear modeling will be used to determine how community characteristics relate to drug use and how these relationships may differ in ethnic minority communities. The second phase of the study will determine, for the first time, how community characteristics relate to change in drug use rates in rural communities and whether there are differences related to ethnic minority communities. The study will involve two waves of re-surveying, one three years and one five years after the initial survey. Resurveyed will be the 30 African American and 30 Mexican American communities and a matched sample of 30 non-minority communities. The surveys and survey methods are identical for all waves. Hierarchical linear modeling will be used to determine how characteristics of these rural communities related to changes in drug use over time.